核心问题
Is the policy trying to correct a market failure, gain bargaining power, protect workers, secure supply, or shift rents across borders?
Trade policy is macro policy when it changes the price system, supply security, regional labor markets, and the bargaining position of firms and states.
The tool is never just a tariff line. It is a choice about who pays, who adjusts, and which supply chain becomes worth defending.
政策目标
Is the policy trying to correct a market failure, gain bargaining power, protect workers, secure supply, or shift rents across borders?
Trade policy changes prices, supply chains, market access, bargaining power, and distribution across sectors before it changes aggregate output.
Trade tools can protect a sector while raising input costs elsewhere. The incidence rarely stops at the border.
工具与手段
政策真正可以调动的杠杆。每种工具都有各自的操作目标、信息需求和政治成本。
Tariffs and quotas
Border measures raise or limit import access and shift incidence across consumers, firms, and foreign exporters.
Trade agreements
Agreements lower barriers while setting rules for investment, services, standards, and disputes.
Export controls
Controls restrict strategic goods, technology, or inputs where national security dominates ordinary market access.
Rules of origin
Content rules shape where firms source, assemble, and document production.
传导渠道
政策工具影响价格、产出、就业、信贷、对外收支或产能的渠道。
进口价格与供应链
除非外国卖家自行降价,关税会抬高进口商品的价格。2018年的证据显示传导是完全的:被征税的进口商品价格没有下降,因此关税落在了美国买家身上;随着部分进口品种消失,企业重组了供应链。
报复与出口需求
贸易伙伴会以自己的关税回应,成本落在与最初争端毫无关系的出口商身上。2018年的关税及其引发的报复,使美国的进口和出口都大幅下降,边境两侧的需求都因此流失。
规则的不确定性
企业决定在哪里建厂、向谁采购,依据的是它们对投资整个周期内关税的预期,而不是当下的税率表。当税率通过总统公告改变、随后又在法院受到挑战时,就像2025年和2026年那样,不确定性本身就可能推迟投资和供应商的更换。
时滞
效果究竟何时显现。有些价格会立即调整;数量、合同、招聘和投资通常反应得更晚。
Price incidence
Weeks to months. Importers, retailers, and exporters split the burden depending on market power.
Supply-chain rerouting
Quarters to years. Firms need contracts, logistics, compliance, and sometimes new factories.
Capital reallocation
Years. A trade barrier does not instantly rebuild skills, supplier networks, or regional capacity.
制度约束
决定一项技术上自洽的政策能否实际使用的法律、政治、资产负债表和操作层面的限制。
Trade law
WTO rules, trade agreements, domestic statute, and retaliation risk bound what can be done.
Input dependence
Protection for one industry can raise costs for downstream firms that use imported inputs.
Distribution
Benefits concentrate in protected sectors while consumer and input costs spread widely.
失灵方式
政策可能失灵、来得太晚或作用于错误边际的各种方式。
Consumer-price leakage
A tariff can look like industrial policy while operating as a regressive consumption tax.
Retaliation spiral
Trading partners can erase the expected gain by targeting exporters or strategic sectors.
Protecting without upgrading
Shelter can preserve existing capacity without building productivity or resilience.
取舍
任何有用的干预都会牵动别的东西。
保护一个行业,还是保护向它采购的行业
对钢铁或半导体征收关税,能帮助这些商品的国内生产者,却会抬高所有把它们用作投入品的企业的成本。一个行业卖给其他行业的产品越多,成本扩散得就越广,这正是评估关税时离不开投入产出表的原因。
收益集中,成本分散
保护带来的好处归受保护行业的业主和工人,他们的人数很容易统计。成本则以更高价格的形式分摊到所有买家身上:一项研究估计,2018年购买进口商品的美国消费者和企业损失了510亿美元,相当于GDP的0.27%,这还未计入关税收入和国内生产者的收益。
成文法,还是行政裁量
设定关税的权力属于国会,国会通过若干法律将其中一部分授权出去,每部法律都附有条件,例如国家安全认定、不公平贸易做法调查或150天的期限。行政部门快速行动在紧急情况或谈判中有用,但2026年2月的裁决表明,依据错误法律征收的关税,即使已经收取,也可能被判无效。
需要监测的数据
关闭本页之后,继续追踪那些可能推翻这一诊断的数据序列。
- Import pricesImport price dataImport prices reveal where tariffs, exchange rates, and supply stress enter inflation.
- Trade balanceTrade balanceDeficits and surpluses must be read with capital flows and domestic saving.
- Sector jobsManufacturing employmentTrade policy should be checked against the sectors and regions it claims to help.
历史案例
检验这项政策诊断的真实案例。
1930
Smoot-Hawley tariff
A downturn-era tariff became the canonical warning about escalation and retaliation.
2001
China joins the WTO
A market-access shock reshaped goods prices, manufacturing geography, and regional labor markets.
2018-2020
U.S.-China tariff wave
Modern tariffs tested incidence, retaliation, and supply-chain relocation in real time.
2025-2026
Emergency tariffs and the Supreme Court
Broad tariffs imposed in 2025 under an emergency-powers statute were struck down on February 20, 2026, when the Supreme Court held that the statute does not authorize tariffs.
Incidence is not the tariff schedule
A tariff schedule names the legal rate. It does not tell you who pays. Incidence depends on market power, contracts, exchange-rate movement, inventory timing, substitution possibilities, and whether foreign exporters cut margins to preserve market share. The WTO reviews every member's trade policies to make them transparent, the largest traders every three years [1], but a schedule alone is never an incidence estimate.
The clearest modern test came from the 2018 U.S. tariff wave. Amiti, Redding, and Weinstein found that the full incidence of the tariffs fell on U.S. importers and consumers, with a reduction in real income of roughly 1.4 billion dollars per month by the end of 2018 [2]. Fajgelbaum, Goldberg, Kennedy, and Khandelwal estimated consumer and firm losses from the full tariff-and-retaliation episode at 51 billion dollars, or 0.27 percent of GDP, before accounting for tariff revenue and producer gains [3]. The analyst's first test is empirical: import prices, consumer prices, firm margins, and sector output should be read together. A tariff can look protective at the border while acting as a cost shock for downstream firms that use imported inputs.
Protection without upgrading is just shelter
The strongest industrial argument for trade protection is not that imports are bad. It is that a sector may need time to learn, scale, coordinate suppliers, or solve a security constraint. That argument only survives if the protected sector faces performance tests. International rules also bound what support can look like: the WTO's Agreement on Subsidies and Countervailing Measures disciplines subsidies and lets members use countervailing duties to offset injury caused by subsidized imports [4].
Export discipline, competition, sunset rules, procurement standards, and measurable productivity goals separate capacity building from a permanent transfer to incumbents. The adjustment that protection is supposed to buy time for is slow. Research on the China trade shock found that local labor markets exposed to import competition adjusted remarkably slowly, with wages and participation depressed and unemployment elevated for at least a decade, while offsetting employment gains in other industries had yet to appear [5]. A temporary tariff that expires before that adjustment is done buys little; a permanent one stops being temporary support.
Trade policy is a blunt stabilization tool
Tariffs and quotas are poor cycle stabilizers. They arrive through legal and administrative channels, change relative prices unevenly, invite retaliation, and can raise inflation when the macro problem is already supply constrained. The 2018 episode showed the side effects: import and retaliatory tariffs caused large declines in both U.S. imports and exports, and the prices of targeted imports did not fall, so the duties passed through in full to U.S. buyers [6].
Trade policy can matter for resilience and bargaining power, but short-run demand management belongs first to monetary, fiscal, credit, and labor-market tools.
The 2025 emergency tariffs and the February 2026 ruling
In 2025 the President imposed tariffs under the International Emergency Economic Powers Act (IEEPA), a statute that, in its half century of existence, no earlier President had used to impose tariffs. One set, aimed at the flow of illegal drugs, put a 25 percent duty on most imports from Canada and Mexico and 10 percent on most imports from China. A second set, aimed at trade deficits and called reciprocal, put a duty of at least 10 percent on imports from all trading partners, with higher rates for dozens of countries [7].
On February 20, 2026, in Learning Resources, Inc. v. Trump, decided together with Trump v. V.O.S. Selections, the Supreme Court held by 6 votes to 3 that IEEPA does not authorize the President to impose tariffs. The Court started from the Constitution, which gives Congress the power to lay duties, and observed that when Congress grants the power to impose tariffs, it does so clearly and with careful constraints [7].
Tariff policy then moved to other statutes, several of which the dissenting justices had listed. On the day of the ruling the President proclaimed a temporary import surcharge of 10 percent under Section 122 of the Trade Act of 1974, which allows a surcharge of up to 15 percent for no more than 150 days unless Congress extends it; it took effect on February 24 and was set to run until July 24, 2026 [8]. Later 2026 tariffs rested on Section 338 of the Tariff Act of 1930, used against Canadian goods in July [9], and on the national-security provision, Section 232 of the Trade Expansion Act of 1962, used for robotics and industrial machinery in October [10].
The ruling also raised the question of refunds. After it, the Court of International Trade ordered Customs to process the affected entries without the IEEPA duties and to refund them with interest. In a notice published on October 5, 2026, U.S. Customs and Border Protection put the IEEPA duties collected between February 3, 2025 and February 24, 2026 at an estimated $166 billion across more than 53 million entry summaries, and described the automated tool, called CAPE, through which importers claim refunds [11]. The economic lesson sits beside the legal one: which statute a tariff rests on decides how fast it can be imposed, how long it can last, and how it can be undone.
情景 · 解读政策的方法
在套用政策框架之前,先核对工具、渠道、时滞和失灵方式
工具
在判断政策立场之前,先说清具体使用的是哪种工具。同一个目标可以通过利率、支出、税收、监管、沟通或资产负债表操作来实现。
传导
追踪政策变动如何经由家庭、企业、银行、市场、预期、汇率和公共部门资产负债表传导。
时滞
区分宣布、实施、市场反应、实体经济反应以及数据发布的时间。
失灵方式
说明哪些情况会让政策适得其反、生效太晚、效果外溢到国外、只是转移风险,或无法触及受约束的部门。